Eradication of buffel grass (Cenchrus ciliaris) on Airlie Island, Pilbara Coast, Western

I. R. Dixon, K. W. Dixon, and M. Barrett Kings Park and Botanic Garden, West Perth, Western Australia, 6005.

Abstract The aims of this project, now into the second year of the implementation phase, are to eradicate buffel grass (Cenchrus ciliaris) from the 25 ha Airlie Island off the Pilbara Coast, to develop and implement methods to restore the indigenous vegetation, and to collect and store seed for future restoration works. The most effective herbicides trialed under these conditions were Roundup Biactive 8 l/ha and Verdict 6 l/ha. Extensive field trials indicated the main perennial shrubby species on the island (Acacia bivenosa, A. coriacea and Rhagodia preissii) are resistant to Roundup Biactive and Verdict. With the exception of native grasses, these herbicides had no adverse effect on other indigenous species. Initial blanket and spot spraying with Roundup to kill the parent followed by blanket spraying, avoiding native grasses, with Verdict, is the most cost-effective regimen for control. A temporary (three years) water pipe for filling battery operated 250 l spraying units was installed across the centre of the buffel populations. Hoses 60 m long with hand held lances were used to apply herbicide. Four operators with two units can spray about two hectares each day. The best time for spraying is six weeks after heavy rain when the parent plants are actively growing and the new seedlings have grown sufficiently to spray. Spraying too early misses most of the seedlings; too late and the seedlings as well as the parent plants are seeding or too senesced to respond to the herbicide. The window of opportu- nity for spraying under these conditions is therefore only two weeks. Results of the spraying on Airlie Island indicate that 98% of the original stands of buffel grass has been controlled. Replanting with greenstock is preferable after heavy rainfall, the main shrubby species planted after spraying with Roundup can then be oversprayed, when required, with Verdict. Greenstock survival rates vary considerably between 5% and 90%, and are entirely dependent on follow-up rainfall. Two to four sprays a year, depending on rainfall events, are required for a period of at least three years (estimated age of soil seed bank) to control this weed with follow up monitoring and backpack spot spraying or hand removal. , a perennial dominant native grass, is best planted at the conclusion of the three year spraying programme to avoid spray damage and for ease of operations to control buffel grass.

Keywords buffel grass control; herbicide; restoration; marine.

INTRODUCTION Buffel grass favours alkaline soils (Christy and Moorby 1975; Griffin 1993). Within the arid zone it establishes Buffel grass in Australia best on areas of higher nutrients and moisture, especially creeklines and floodways. Buffel grass (Cenchrus ciliaris) is a perennial grass native to , the Middle East and southern . It was first There are several buffel grass varieties in Australia, each introduced to Australia in the packsaddles of Afghan camel with different growth habits and requirements. Seed dor- drivers (Bryant 1962) and was later used by the pastoral mancy and germination characteristics may also be vari- industry for erosion control and as a pasture supplement able between or even within varieties. The varieties biloela, throughout the Pilbara and Kimberley regions of Western gayndah and others are African in origin and are widely Australia. This grass has also established on a number of favoured as pasture feed in Queensland. The Western Aus- islands off the Pilbara coast, including Airlie Island. tralian (WA) variety is shorter, reaching a maximum of 75 cm and not as vigorous (Humphrys 1974). Curiously, seed- Introductions of buffel grass were primarily aimed at im- lings of the WA variety have lower drought tolerance, but proving stockfeed, stabilising soil and revegetating bare flowers much more quickly following rains, and is hence and eroded areas (Bryant 1962; Humphrys 1974). The able to survive and spread after rare rainfall events. The effectiveness of buffel grass at stabilising soils is due to WA variety was originally introduced from the Middle the ready germination, rapid propagation and easy estab- East, and has possibly since become further adapted to lishment, even on bare or infertile soils (Bryant 1962). arid Australian conditions. Buffel grass is resistant to drought, fire and heavy graz- ing, so it is dominant and very persistent at a site once In higher rainfall areas of Queensland, buffel grass does established making it useful as an arid zone pasture grass not spread rapidly, if at all, and usually requires cultiva- (Bryant 1962; Hodgkinson et al. 1989). These character- tion to establish a population (Hacker and Ratcliff 1989). istics are attributed to the robust root system and swollen In the more arid conditions in central and western Aus- stem bases, which accumulate carbohydrate reserves, so tralia, however, buffel grass (WA variety) is much more that loss of leaf lamina during drought or after fire is not invasive. Its resistance to fire, drought and grazing make fatal to the plant. Regrowth may then be rapid in favour- it extremely persistent, and its rapid growth and flowering able conditions (Humphrys 1974). allow it to dominate over native vegetation in some areas

Pages 92-101 In Veitch, C. R. and Clout, M. N. (eds.). Turning the tide: the eradication of invasive species. IUCN SSC Invasive Species Specialist Group. IUCN, Gland, Switzerland and Cambridge, UK. Dixon et al.: Eradication of buffel grass on Airlie Island

(Humphries et al. 1991). The major mechanisms of dis- and by 1993 had spread over 2.2 ha (WMC 1993). By the persal are wind, flood, fire (Griffin 1993) and possibly year 2000 it covered an estimated area of 8 ha. domestic stock. Seed is also easily spread by humans as they readily adhere to trousers and socks, thus it is very The initial spread of buffel was relatively slow, however important to make sure seeds are removed from clothing disturbance events appear to have enhanced its invasive after visiting a buffel grass area. The spiked seed-bearing capacity. The proposed decommissioning of the Airlie involucre also increases spread by attaching to animals. Island installation presents another disturbance event which Vehicle wind-assisted spread along roads is also evident could allow buffel infestation to further increase. in Uluru National Park (Griffin 1993). Buffel grass was not reported as spreading until the 1970s, when high rain- On Airlie Island the presence of buffel grass threatens the fall and floods lead to rapid colonisation along creeklines natural plant communities by replacing, almost entirely, and alluvial flats (Griffin 1993). the understorey cover of indigenous grasses and herbs. Buffel grass has already become a dominant plant species The biology of buffel grass allows populations to be self- on the island and other islands along the Pilbara coast. maintaining, and encourages rapid spread in favourable This environmental weed substantially increases the fire conditions. As a result, buffel grass (WA variety) is an risk which may impact upon the habitat of local fauna and aggressive coloniser of native habitats, especially moist may cause significant and permanent changes in vegeta- environments, where it forms dense monocultures, exclud- tion structure and diversity. ing other species (Humphries et al. 1991). Buffel grass also alters fire regimes by increasing fire frequency and At the start of spraying in 1999, buffel grass formed a near intensity, and while buffel survives, native species are sup- monospecific stand over eight hectares (33%) of the is- pressed or replaced (Griffin 1993; Humphries et al. 1991). land. This weed may release allelopathic chemicals (Choo 1984) into the soil that inhibit growth of other species, The success of buffel grass raises serious concerns for the potentially acting as a key displacement agent for most of welfare of plant and animal species that are restricted to the native vegetation. Buffel grass may also be detrimen- moist sites throughout the arid zone (Humphries et al. tal to the island’s fauna, especially the breeding cycle of 1991). These habitats are critical refugia for survival of shearwaters and the survival of herpetofauna. numerous plants and animals (Griffin 1993; Humphries et al. 1991; ANCA 1996). Urgent control methods are re- Options for controlling buffel grass quired in central Australia (Humphries et al. 1991), espe- cially in national parks and nature reserves. A number of options were canvassed at the outset of the study including biological control which was deemed in- Buffel grass has been reported as a weed or a serious in- appropriate because of potential adverse impact on the vader in Western Australia in the following reserves and pastoral industry throughout northern Australia. Fire was biologically significant areas: Cape Range National Park, ruled out as a control measure as buffel survives fire (Grif- western coastal plain of the Carnarvon-Exmouth area and fin 1993; Humphries 1991) and there is a complete fire Doole and Roberts Islands where it is potentially a threat. ban on the island due to the risk associated with the oil Numerous other islands in the Shark Bay - Exmouth area storage facilities as well as impacts on the indigenous flora are exposed to invasion by buffel grass (Department of and fauna. Although isolated plants on the island are pulled Environment, Sport and Territories 1996). up, physical removal is not usually appropriate due to the large number of plants and the difficulty of removal be- Buffel grass in Western Australia cause of their strong root system. Other problems are the cost, soil disturbance and possible wind erosion (blow out In 1910 the first deliberate introduction of buffel grass in from cyclonic wind). Mowing is ineffective, as well as Western Australia was carried out on Wallal Station, West- costly, impractical (petrol mowers cannot be used due to ern Australia. Since then numerous introductions have fire risk) and possibly damaging to native fauna. The key been made, for example aerial sown seed on option for research into the control of buffel grass focussed Mundabullangana Station between 1926-1928 (Bryant on herbicide control. 1962). Key aims of the project Buffel grass seed was most likely brought to Airlie Island in soil used in the construction of the lighthouse in 1913 Phase one of this project was to investigate and research (WMC 1993). By 1987, buffel grass roughly occupied a the biology of buffel grass and to develop a control pro- 2.5 ha kidney-shaped area around the lighthouse (Astron gramme which will integrate eradication or sustainable 1988). Later that year Western Mining Corporation con- control of buffel grass with the reinstatement of indigenous structed an oil installation on the island, removing approxi- species. mately 1.5 ha of buffel grass in the process (Astron 1988). Phase two of the project is the control of buffel grass over Following construction of the oil installation, buffel grass the whole island based on the results of Phase 1, restora- was estimated to be spreading at a rate of 0.2 ha per year tion of indigenous communities, and initiation of a seed

93 Turning the tide: the eradication of invasive species collection and storage programme for future revegetation The island has been used in the last decade as a fuel stor- works. This phase has been underway for two years, and age depot for Western Mining Corporation Petroleum Di- is the emphasis of this paper. vision (WMC), Novus Petroleum’s offshore oilfield projects and by the present lease holder Apache Energy. Outcomes of this study will be directly relevant to the Imminent decommissioning of the petroleum activities on decommissioning of the Airlie Island oil installation while the island require that buffel grass control be undertaken providing benchmark data on the control of buffel grass. to ensure that the weed is contained, controlled or eradi- The study will therefore be of regional, national and inter- cated. Though the lessee is only required under agree- national significance to land managers and conservation ment to control buffel grass and revegetate in the leased agencies where buffel grass is an environmental weed. area (approximately half of the island), they are undertak- Information on buffel grass control generated from this ing a programme to try to eradicate buffel grass from the programme is already being utilised by local land manag- entire island and revegetate areas, where necessary. ers on adjacent islands and adjacent mainland as well as other areas in Australia such as Queensland and central Airlie Island is on the borders of the Fortescue and Australia. Carnarvon Botanical districts (WMC 1988), and receives an annual rainfall of 300 mm from both summer (cyclonic) Data was generated in Phase one of the project including and winter (cold front) storm systems. Much of this rain buffel grass seed production/viability; longevity of the soil falls intensively in irregular intervals, often accompanied seed bank (at least three years); buffel spread; densification by severe winds. The soil on the island is white to pink- and percentage groundcover; seedling recruitment; soil nu- ish, generally coarse, calcareous sand with poor water trient profiles; indigenous plant resistance to herbicides holding capacity. (8 l/ha Roundup no effect on Acacia bivenosa, A coriacea or Rhagodia); life cycle of buffel (seed can germinate and The rainfall pattern, lack of permanent surface water and flower in six weeks, plants can grow, germinate, flower or small size of the island means the vegetation must be seed any time of the year as long as conditions are favour- drought tolerant and able to cope with salt-laden wind. able, a minimum of 10mm rainfall) and comparison with Most of the island is dominated by two Acacia species (A. Eulalia aurea the dominant indigenous grass species. Phase bivenosa and A. coriacea), with Rhagodia preissii, Eulalia one also found that repeated spray trials (Roundup then aurea and now buffel grass (Cenchrus ciliaris) as the ma- Verdict) after regrowth gave excellent kill rates as did seed- jor perennials. Shorelines are dominated by Spinifex ling spray trials. Detailed summaries of the results of re- longifolius, Ipomoea pes-caprae, Sporobolus virginicus search undertaken in Phase one are being prepared for pub- and Eulalia aurea. During favourable seasons, a large lication. number of annual species can be found in abundance, in- cluding Portulaca intraterranea, Euphorbia spp., Study site Boerhavia repleta, Cleome viscosa, Cuscuta australis and Threlkeldia diffusa (WMC 1988). Airlie Island (Fig. 1) is a 25 ha nature reserve and lies 35 km north-east of Onslow. Only two other weed species are known to exist on Airlie Island. Cotton bush (Aerva javanica) plants may be found anywhere on the island. When found during spray opera- tions these are pulled up and any fallen seed is picked off the ground; plants are placed in sealed polythene and taken off the island with all other rubbish for disposal (deep bur- ied). Any plants seen by Apache Energy staff are treated in the same manner. Only a few plants are found each year; locations of these plants are noted and occasionally these sites are visited by Apache staff to check for new plants. The other weed is a native, Abutilon lepidium, that occurs naturally on adjacent islands but not on Airlie. To date about 300 seedlings have been removed from one small area (Long pers. comm.). METHODS

Aerial Photography and Image Enhancement

Aerial photographs of Airlie Island are taken annually. Fig. 1 Airlie Island with infrastructure, lease Visual examination of these images from 1993 to 1998 boundary and approximate buffel grass area shows only a few small clumps of buffel grass colonising when eradication work began. away from the main buffel zone. There are numerous small

94 Dixon et al.: Eradication of buffel grass on Airlie Island plants known to exist that do not show up on the photos. the toxicity of pesticides) and widely available to the gen- The major feature of the series of images is densification eral public, there was no other formulation available, to and gap-filling within the existing stand, which is respon- our knowledge, that would remain evident for several days sible for most of the increase in buffel grass cover on the under the hot spraying conditions experienced on Airlie. island. We also decided not to use any spraying oils for fear of damaging foliage in the high temperatures and thus inhib- The set of photographs taken in June 1997 were scanned iting the translocation of herbicide into the buffel plants. and colour enhanced for the wavelengths corresponding to buffel grass, beach spinifex and Eulalia. These areas Spray plots were coloured yellow, purple and green respectively to enhance visualisation of the extent of buffel grass spread. Spray plot sizes were 1x1 m, each separated by a 0.5 m- wide corridor. Three replicas of each plot were made Selection of herbicides within each trial site. All plots were sprayed working sys- tematically across the plot and back again in the opposite The key aim of the research programme was to identify a direction to ensure even coverage. Compressed sheeting safe and effective herbicide to control buffel grass with was used as a barrier to ensure there was no drift onto minimal impact on the indigenous flora and fauna. Buffel other plots. grass was known to be very difficult to control due to its ability to survive long periods of drought and resprout from Trial I dormant buds. Other species with similar traits (e.g., Condition of plants before spraying was scored using Ta- calycina, Eragrostis curvula and Hyparrhenia ble 5 based on the experience of the authors. hirta) can be successfully controlled using the correct her- bicides and spraying at the right stage of growth. For ex- Sprayed in February 1997 when the plants were showing ample, spraying Ehrharta calycina before flowering re- vigorous new growth after the summer wet season, but sults in a high death rate, whereas spraying at flowering showed some signs of drying out (spraying condition 4 time results in dieback of the top part of the plant but going on 6). resprouting from dormant buds at the base (Dixon 1999). Herbicides known to control other Cenchrus species (e.g. Treatments used were: Fusilade 212, 2 l/ha (plus Agral 60 Cenchrus echinatus), and which would be suitable for @ 1 ml/l); Fusilade 212, 4 l/ha (plus Agral 60 @ 1 ml/l); Airlie Island conditions, are the grass-selective herbicides Fusilade WG, 5g/l @ 2 l/ha (plus Agral 60 @ 1 ml/l); Fusilade 212 and Verdict 104 (Parsons 1995). Fusilade WG, 10g/l @ 4 l/ha (plus Agral 60 @ 1 ml/l); Roundup Biactive, 3 l/ha; Roundup Biactive, 6 l/ha; Con- Successful trials and large-scale herbicide application on trol. other grasses have been conducted in Kings Park bushland, Perth, Western Australia, indicating several of the grass- Trial II selective herbicides are safe to use over most indigenous Sprayed in October 1997 when the weather was warm and plants including such as sedges and the buffel grass showing signs of drought stress. The buffel rushes, kangaroo paws (Haemoderaceae) and orchids, and grass was drying out, and had only a few obvious green dicotyledons such as the Proteaceae family. As we had shoots (Spraying condition 6, Table 5). extensive experience with the following herbicides, the first trials included the grass selective herbicides Fusilade 212® Treatments used were: Roundup Biactive 0.6,1.5 and 3 (212g/l fluazifop-p) Fusilade WG® (212 g/kg fluazifop- l/ha; Targa 6 and 8 l/ha + Agral 60 @ 1 ml/l; Verdict 6 and p) (this formulation of wettable granules is easier and safer 8 l/ha + Agral 60 @ 1 ml/l; Fusilade 212, 6 and 8 l/ha + to handle and does not smell of hydrocarbons), Targa® Agral 60 @ 1 ml/l; Control. (99.5 g/l quizalofop-p-ethyl) and the non-selective herbi- cide Roundup Biactive® (360 g/l glyphosate) (the wet- Trial III ting agent in this formulation is claimed to be safer to use Sprayed in April 1998 in low to moderate wind condi- over fauna, particularly frogs). It should be noted tions. Plants were responding to recent heavy rain with Roundup® 360 (360 g/l glyphosate) and other glyphosate vigorous new growth, in spraying condition 3 (Table 5). formulations had been tried by several people to control This first rain was followed by regular rainfall events for buffel grass in central and eastern Australia, their results the following 11 weeks. were very poor and we were advised not to use it. Ver- dict® 104 (104 g/l haloxyfop) was included in the latter Treatments used were: Fusilade 212, 6 and 8 l/ha + Agral trials. These herbicides were evaluated over a three year 60 @ 3 ml/l; Targa 6 and 8 l/ha + Agral 60 @ 3 ml/l; period, not all the trials will be included in this paper. Verdict 6 and 8 l/ha + Agral 60 @ 3 ml/l; Roundup Biactive, 0.6, 1.5 and 3 l/ha; Control. Redeye®, a marker dye incorporated in the spraying tank, NOTE: The rate of Agral 60 was altered from the previ- was used in the larger trials to make sure there was no area ous two trials (3 ml/l up from 1 ml/l) following the manu- missed and over-spraying was avoided. Though Redeye facturer’s recommendations. is a Schedule 6 poison (Australia-wide schedule outlining

95 Turning the tide: the eradication of invasive species

Large-scale spraying trials Equipment used A few large-scale (100m2) trials were attempted with her- 2 x 250 litre sprayers with 12 volt battery operated pump; bicides achieving a high kill-rate in the multi-herbicide 2 x 60 metre reel hoses and adjustable spray guns; 1 x 12 trials. These trials using a 15 litre back-pack sprayer were volt battery charger; 6 x 12 volt car batteries; 600 metres x aimed at confirming the effectiveness of herbicides sprayed 32mm ID pipeline; 10 taps miscellaneous joiners/fittings; on a larger scale prior to the implementation phase. 2 Hardie backpack sprayers.

Roundup Trials Full-scale spraying operation Three large areas were sprayed in April 1998 with 6 l/ha Roundup Biactive at 8 l/ha and Verdict at 6 l/ha were used. Roundup Biactive as a broadscale trial. Plants were in The strategy employed Roundup in the first spraying op- spraying condition 3 (Table 5). eration as there were few annual indigenous plants emerged that might be affected by the spray. Then Verdict used as Further trials were sprayed at 8 l/ha under spraying condi- a blanket spray as annuals and possibly some native per- tion 4 (Table 5). ennial plants are at a susceptible stage.

Verdict Trial Each spraying unit requires two operators, one as a sprayer A large area around the buildings was sprayed in July 1998 and the other to release and withdraw the hose. Operators with Verdict at 8 l/ha. Plants were in spraying condition 4 needed to change duties on a regular basis to avoid fatigue (Table 5). as a result of the high temperatures. Operators avoid spray drift by spraying upwind which can be difficult as you have Implementation phase to spray around bushes and under them whilst avoiding spray application of the bushes. In dense stands we use a This phase involved spraying all of the buffel grass on the blanket spray and other areas spot spray. Spray is applied island with approved, effective equipment. Some of the in a circular fashion, pulling the hose out and spraying as key constraints which we had to address are as follows: you go to the extremities of the hose, then working back- wards to the spray unit and again pulling the hose out until v Getting the equipment on the island. This was trucked the entire circle is completed. The unit is then moved onto up to the remote townsite of Onslow then placed on a the next section. Some areas not covered have an extra barge which goes to the island once a week. No large, section of hose added to cover the area; alternatively we heavy or dangerous items can be carried on the heli- use a backpack sprayer for outliers or remove isolated copters which take personnel to the island from Bar- plants by hand. Most of the leased area of the island can row Island, the direct flight destination from Perth. be accessed by using a road or around the bund surround- v Using the best equipment for the job and making sure ing the oil tanks. In these areas we placed the spraying we had adequate spare parts. No internal combustion tank on a tractor for ease of movement. engines are allowed on the island (due to possible hy- drocarbon contamination/spills), therefore we had to use Round One Spraying - March 1999 battery-operated spraying units. Water supply for spray- The first round used Roundup for ease of operations and ing is from a reverse osmosis supply generated on the because it is more cost effective than using Verdict. Her- island, rainwater also goes into the tanks. This supply bicide application was following six weeks of heavy soak- is limited; on one occasion the pipe to the tanks burst ing rain. Ideally the spraying programme should have been and we only had just enough water for spraying opera- initiated two weeks earlier as most of the buffel grass had tions. just reached first anthesis and early seeding. The buffel v Cyclones during spraying operations. When these de- grass was in rapid growth with some plants beginning to velop the island is evacuated and all operations ceased. dry out by the fourth day of spraying. Most of the applica- v Laying a temporary (three years) water pipe with taps tion was blanket spraying of heavily-infested buffel areas, for filling the spray tanks. avoiding as many indigenous plants as possible with very v The 250 litre spray tanks have to be carried to each few annual seedlings present. Previously-sprayed areas station by hand, no vehicles allowed in the natural veg- were spot sprayed, avoiding contact of Verdict on Eulalia etation areas. It is therefore necessary to judge very to reduce damage to planted greenstock and annual indig- carefully how much material you need in the tank to enous plants. Though original trials indicated Eulalia aurea finish off the area. was resistant to the grass-selective herbicides, probably v Coping with the extreme weather conditions (e.g. 36°C because they were under stress at the time, subsequent tri- and 80% humidity during summer spraying operations) als showed they were very sensitive and future spraying and difficult working conditions (e.g. walking back- would need to avoid excessive contact with Eulalia. wards when spraying) and abundant dead twigs which fouled boots and clothing. Round Two Spraying - June 1999 v Training for the job and safety issues and the need for The second round of spraying was mainly blanket spray- annual renewal of Helicopter Underwater Escape Train- ing resprouts and seedlings with Roundup, as there was ing (HUET). such a large amount to spray and low levels of annual in- digenous plants in the previously-sprayed area. Some spot

96 Dixon et al.: Eradication of buffel grass on Airlie Island spraying with Roundup in outlier areas and spot spraying revegetation of bare areas. This density was doubled to Verdict over replanted areas was done. allow for assumed seedling death. The total area requir- ing revegetation is estimated at one hectare; this excludes Round Three Spraying - April 2000 the leased area which is to be revegetated after The third round of spraying followed six weeks of sub- decommissioning the oil facilities. stantial rainfall on the island as a result of a cyclone. The buffel was in excellent condition for spraying. About half Planting Times of the area, low impact areas, were sprayed with Roundup; The lack of summer rain resulted in plantings only being the remainder, rich herb fields, with Verdict. Backpack undertaken in winter. sprayers were used on outlier populations. A thorough inspection after the spraying operation detected some other As soon as the plants arrived on the island, they were placed plants which were removed by hand. in a shady position and watered. Planting was done with a garden trowel, the root balls of the plants were placed just Revegetation below the soil surface to avoid drying out. The first plant- ing trial in April 1998 consisted of A. bivenosa, A. coriacea, Greenstock propagation Rhagodia preissii and Eulalia aurea which were planted All seed was collected from Airlie Island to ensure only into moist soil. A further trial in June/July used only 200 local provenances were utilised. Plants of A. coriacea, A. A. bivenosa. A large trial in June 1999 used 1400 plants: bivenosa, Rhagodia preissii and Eulalia aurea were propa- A. bivenosa and E. aurea in high numbers; A. coriacea gated from seed in glasshouses at Kings Park and Botanic and R. preissii in low numbers. These were planted in Garden and transported to Airlie Island. Acacia seed were very dry conditions. Planting was in a number of loca- hot-water treated and left to soak overnight. Seed was tions in areas previously heavily-infested with buffel grass. sown into punnets and after about six weeks, seedlings After cleaning with bleach and thoroughly rinsing, the were pricked out into tubes. Seed was sown in summer spraying units and long hoses were utilised to water the for planting in early winter of the following year. Propa- plants in. gation of larger seedling numbers was in deeper 5cm x 5cm x12cm pots to promote stronger root development RESULTS and to give a longer holding period in case of dry condi- tions. In 1999 plants were raised in an ‘Accredited’ (ap- Aerial Photography and Image proved by the Nursery Industry Association of Western Enhancement Australia) commercial nursery, which was inspected be- forehand and during the growing period to ensure adequate Aerial photographs taken in June 1997 were scanned and hygiene and weed-free propagation. Using an accredited colour enhanced for the wavelengths corresponding to nursery as the source of all plant material reduces the risk buffel grass, beach spinifex and Eulalia aurea. From this of introducing pests, diseases and other environmental it was estimated that 6 ha, or 23% of the island was in- weeds. fested with buffel, and that there was virtually no overlap in the distribution of buffel grass and Eulalia aurea. Fur- Transport of the seedlings to the site was by truck to ther data collected in 2000 indicated that buffel grass ac- Onslow, then barge to Airlie Island. Plants were watered tually covered almost 8 ha of the island. prior to packing and transport to the trucking company. Polystyrene foam boxes and strong waxed cardboard boxes Spray Plots were used to transport the seedlings.

Replanting density Trial I Benchmarking (using quadrats and transects) in pristine areas of the island gave a figure of natural plant density The trial was scored 11 weeks after spraying (Table 1) and and the estimated number of plants required for assessed again after six months. This later assessment, after some resprouting from dormant buds, found no dead

Table 1 Spraying Trial I; Concentration and Percentage Percentage applied in February 1997. Herbicide application rate live plants death of above flowering ground biomass

Roundup Biactive 7.5ml/l @ 3 l/ha <40% 88% ± 7% 15ml/l @ 6 l/ha 55% 88%± 7% Fusilade 212 5ml/l @ 2 l/ha 65% 4%± 4% 10ml/l @ 4 l/ha 68% 22%± 12% Fusilade (WG) 5g/l @ 2 l/ha 77% 0 % 10g/l @ 4 l/ha 53% 23%± 20% Control — 100% 0 %

97 Turning the tide: the eradication of invasive species plants in the control or any Fusilade plots (except one liq- Table 3 Effectiveness of herbicides on kill of uid application @ 2 l/ha in one plot, in which 20% of the buffel grass in trial 3. plants had died). In contrast, most plants were dead in all Roundup trials. Plants that were still alive were flower- Herbicide Concentration and Percentage ing. A similar situation was found 18 months after spray- application rate death of above- ing, but numerous germinants had emerged in the Roundup ground biomass plots. Roundup 1.5ml/l @ 0.6 l/ha 57% ± 4% A comparison of seeds collected 11 weeks after the trial 3.75ml/l @ 1.5 l/ha 52% ± 10% from sprayed and unsprayed buffel showed that all the flo- 7.5ml/l @ 3 l/ha 67% ± 5% rets of sprayed plants were empty, compared with the ± unsprayed plants, which had 80% of florets with seed, with Fusilade 212 15ml/l @ 6 l/ha 71% 18% an average of 1.50 ± 0.18 seeds per floret. It appears that 20ml/l @ 8 l/ha 100% ± 0% although spraying under non-optimal conditions is not very Verdict 15ml/l @ 6 l/ha 95% ± 5% effective at killing plants, it may prevent viable seed-set, 20ml/l @ 8 l/ha 95% ± 5% and therefore be worthwhile. Targa 15ml/l @ 6 l/ha 44% ± 5% Conclusion 20ml/l @ 8 l/ha 34% ± 23% Both Fusilade formulations had low kill rates (~20%) at Control - 0% the higher concentration of each. Roundup had a better kill-rate of (~80%), with most plants succumbing the fol- Trial III lowing drought period. The dry weather conditions at the time of spraying limited the effectiveness of the herbicides, The trial was scored in July 1998, 10 weeks after spray- although Roundup had achieved complete kill when it was ing. During this time rains had been consistent, with sev- reassessed in August. eral large falls. Conditions for plant regrowth were ex- ceptionally good, so the results in Table 3 are the worst Trial II (i.e. maximum regrowth) that could be expected from a The trial was scored six months after spraying, when the spray applied in good conditions. next rains fell. These results suggest that, under optimal spraying condi- The trial was scored again in July 1998, nine months after tions, Targa is of limited use, Roundup provides moderate spraying. All plants were alive, resprouting from dormant kill-rates, and Fusilade and Verdict both give excellent re- buds. sults, killing nearly all plants (above-ground biomass) sprayed. Examination of larger Roundup trials showed Conclusion that plants resprouted from the spray-shadowed portion, No herbicides caused any buffel plants to die, owing to killing only leaves that were sprayed directly. Verdict and the unfavourable spraying conditions. These results show Fusilade, however, seemed to usually kill the entire clump the importance of spraying plants at the correct stage of (above-ground biomass) in a single application. Verdict growth, as it appears that stressed plants were drying out worked equally well at both concentrations, while Fusilade and resulting in a lack of translocation of the herbicide appeared to work better at the higher concentration, which from the leaves through to the dormant buds at the base of killed (above-ground biomass) all plants sprayed. Some the plant. resprouting of all treatments occurred at a later date but not as much as usual, indicating the plants were sprayed Table 2 Effectiveness of herbicides on kill of under ideal conditions. buffel grass in trial 2. Condition of Plants Herbicide Concentration and Percentage application rate death of above- The three multi-herbicide trials described previously were ground biomass applied under different conditions, and showed a marked variation in effectiveness. The major difference between Roundup 1.5ml/l @ 0.6 l/ha 14% ± 8 % trial conditions was the condition of the plants. The spray- 3.75ml/l @ 1.5 l/ha 0 % ing conditions and results are listed in Table 4. 7.5ml/l @ 3 l/ha 14% ± 14 % Plants that were vigorously growing were killed more ef- Fusilade 212 15ml/l @ 6 l/ha 20 %± 20 % fectively, while senescing plants were not killed, or in the 20ml/l @ 8 l/ha 0 % ± case of Roundup, killed up to the onset of leaf desicca- Verdict 15ml/l @ 6 l/ha 94% 6 % tion. 20ml/l @ 8 l/ha 0 5% Targa 15ml/l @ 6 l/ha 12% ± 6 % Spraying conditions 20ml/l @ 8 l/ha 11% ± 11 % Buffel grass must be actively growing (in spraying condi- Control - 0 % tion 3 to 5, Table 5). Do not spray senescing plants.

98 Dixon et al.: Eradication of buffel grass on Airlie Island

Table 4 Susceptibility of buffel grass to herbicides at different stages of plant condition

Plant Condition* Herbicides Herbicides achieving Herbicides achieving achieving <35% kill 35%-75% kill 80%-100% kill

3. Vigorous new Targa Roundup, Fusilade at Verdict, Fusilade at growth lower concentration higher concentration

4. Vigorous growth Targa Targa, Fusilade at lower Verdict all concentrations, and flowering concentration, Roundup Fusilade at higher at all concentrations concentration

6. Senescing Targa, Verdict, Roundup (if plants just No herbicides achieved Fusilade most starting to senesce) this kill rate conditions

* Full description of plant condition in Table 5 Large scale spraying trials Implementation phase – the eradication of buffel grass on Airlie Island Roundup All areas sprayed at 6 l/ha had a high rate of kill where Round One Spraying - March 1999 Roundup had been directly applied to leaves, but plants Most of the buffel could be seen to be yellowing by the and portions of plants within the spray shadow were fourth day of spraying, indicating a good kill rate. Cover- resprouting. It is therefore necessary to spray a second age appeared complete, but occasional unaffected shoots time when the buffel has had time to resprout. were detected as had been previously observed in herbi- cide trials. The results from the 8 l/ha trials were outstanding with very few resprouts indicating this is the appropriate rate Cyclonic rain shortly after spraying led to an unexpected for use for the implementation programme. rapid regrowth of large plants, however conditions by the end of April were too dry to spray. Desiccation of plants Verdict Trial and the combined effect of the spray prevented seed set on Verdict at 8 l/ha, sprayed around the buildings in July 1998 the resprouting culms. gave excellent results, killing all the seedlings sprayed and most of the parent plants. Some plants missed spray application, as expected, mainly under acacias as well as occasional plants away from the Conclusion main populations. One area which was showing signs of Roundup has a very high kill-rate where it can be applied drying out towards the end of the spraying period exhib- directly to the leaves, but for large plants a second spray is ited reduced death rates. However, overall results were needed after the above ground biomass death and subse- better than anticipated with an estimated 80%-90% kill quent resprouting to destroy plants and culms which were rate of mature plants, though there was resprouting on a in the spray shadow. Verdict is ideal for spraying seed- number of plants. lings and will also kill previously-sprayed resprouting par- ent plants as well as some plants not previously sprayed.

Table 5 Buffel grass plant condition and suitability for spraying

Condition Plant Growth Stage Suitability for spraying Number

1 Fully dormant (no visible green tissue). Plants not growing. DO NOT SPRAY. 2 Bud-break (no extended leaf lamina Not enough leaf area for herbicide yet visible). uptake. DO NOT SPRAY. 3 At least half of shoots with extended Conditions acceptable for spraying. leaf lamina. Not yet flowering. 4 All shoots with extended lamina. Conditions perfect for spraying. Flowering evident. 5 Vigorous shoot growth. All mature Conditions perfect for spraying. culms fully extended. 6 Any evidence of senescence of leaves Plants not growing. DO NOT SPRAY. or culms, i.e. leaf-rolling or desiccation (leaf margins dry).

99 Turning the tide: the eradication of invasive species

The method of spraying, though time consuming, worked Planting well. Spray equipment was adequate and effective with no The results from the first two planting trials were outstand- breakdown, and the batteries lasted longer than predicted ing with 89% to 97% of all species surviving and growing (five hours) before recharging was necessary. The correct vigorously 10 weeks after planting. positioning of the water pipe and taps assisted smooth operation, and judging the appropriate amount of spray to Further monitoring in April 2000 indicated there was no finish spraying an area worked well. On average, four op- further death in the April or June/July 1998 trials. This erators with two units can spray two hectares each day, may have been due to substantial rainfall after planting. or four days to spray the entire buffel-infested area. The large planting in June 1999 was a failure with no rain Round Two Spraying - June 1999 for six months after planting. Almost all the plants of all As the parent plants had decreased significantly in number, species died. Though no exact count was done, a few plants regrowth was easier to target with spray. There was a sig- of A. bivenosa are alive in two or three locations, with nificant increase in the number of seedlings, but not as survival of <1%. high as anticipated in previous trials; this may be due to the drier conditions experienced in 1999. The extent of DISCUSSION the buffel infestation appeared not to have increased or decreased, but the area took less time to cover as most of The research results from Phase one of the programme the main stand of buffel had been decimated. The kill rates underpinned the implementation phase of the programme. were high, eliminating most of the resprouting plants and A good knowledge of the biology of the buffel plants other parent plants that were missed in the first spray. There opened up a narrow window of opportunity for spraying was little damage to non target species and certainly no and optimising kill rates, and seed production, soil seed recorded damage to the perennial species, with the excep- bank and longevity of seed enabled us to plan ahead with tion of some damage to young Eulalia aurea plants in a work schedule. This is aimed at eradication of buffel revegetation trials where Verdict was sprayed. grass from Airlie Island in the near future, as long as the appropriate funding is available. Round Three Spraying - April 2000 The third spraying was mainly of seedlings, as few mature Spraying needs to occur whenever there is sufficient rain- live plants remained. At this stage we estimate most of fall to control the few remaining mature plants and the the seedlings and over 98% of mature buffel grass plants emergent seedlings of buffel grass. As the estimated age on the island were eradicated. Seedlings may germinate of the soil seed bank for buffel grass (Phase 1 unpublished from the soil seed bank. However, the viability of the soil data) is three years, rapidly declining after eighteen months, seed bank is rapidly declining and sustainable control of further large-scale spraying operations will not be neces- buffel grass will depend upon careful monitoring and a sary. Trials with motorised, battery operated backpack judicious ‘mop-up’ spray programme. sprayers indicate this is the best option for spraying smaller numbers of plants. Hand-operated units are too debilitat- Revegetation ing for operators in the hot, humid conditions experienced on the island. Greenstock propagation The seedlings produced both in the small trials and by the The initial trials provided an appreciation of the problems accredited nursery were healthy and vigorous, and no and how adaptable you have to be with your implementa- weeds were present in the pots. The most successful con- tion programme which is entirely dependent on the vagar- tainer for transportation was found to be polystyrene foam ies of the weather. The spraying results proved to be in- boxes as strong waxed cardboard boxes became moist and consistent as every trial was different. However, the re- collapsed, causing some damage to seedlings. sults from the large-scale spraying programme were used as the basis for the implementation phase. The authors Replanting density are confident that the herbicides, and their concentrations, The density of planting in our trials proved successful and used in the implementation phase are ideal for the expected we recommend a planting rate that would result in A. varying conditions of the plants. Future operations, due to bivenosa and A. coriacea at a density of 1250/ha and the recruitment of herbs in the buffel areas, will be using Rhagodia preissii and Eulalia aurea at a density of Verdict to avoid damage to these plants. Planting Eulalia 2500/ha. aurea should be delayed until the soil seed bank is elimi- nated, otherwise further spraying could kill or damage We recommend that planting densities be double to allow Eulalia. for a 50% death rate. However, we must be vigilant as A. coriacea has declined in some areas and it is much slower The best time for spraying adult plants is within the range growing when compared to A. bivenosa which may even- of three to five weeks after sufficient rain has fallen. If tually outcompete it and require thinning. spray is applied too early the seedlings are too small to target and with extra rainfall there may be delayed germi- nation. Also it is not cost-effective to spray twice when

100 Dixon et al.: Eradication of buffel grass on Airlie Island one operation can achieve good results. Under quick dry- Personal communications ing conditions some plants are beginning to get stressed, Long, Vicki (Oct. 2000), Astron Environmental, Karratha, and may occasionally drop some mature seed before they Western Australia. are sprayed. REFERENCES Logistically there are major problems with a remote insu- lar site such as Airlie Island. For example we have to Astron Engineering. 1988. Airlie Island Terminal, first an- rearrange our usual work schedules when the spraying is nual environmental report, May 1988; Unpublished re- necessary as rainfall in the arid zone is erratic. All the port to Western Mining Corporation. accommodation, flights and so on are arranged by the oil company and, if urgent work is being done on the oil in- Australian Nature Conservation Agency. 1996. Environ- stallations, securing necessary flights and accommodation mental weeds causing serious concern - buffel grass; can be problematic. http://www.anca.gov.aau/plants/manageme/ envgrass.htm Though the spraying programme is in hand, the programme Bryant, W. G. 1961. Buffel grass (Cenchrus ciliaris L.) has identified an urgent need to revegetate bare areas. The for erosion control. Journal of Soil Conservation, New dead below-ground biomass of buffel does provide soil- South Wales 17 (3):135-147. binding to prevent wind erosion for at least three years, in which time revegetation needs to proceed. Christie, E. K. and Moorby, J. 1975. Physiological re- sponses of semiarid grasses. I: the influences of phos- Planting greenstock can give excellent results if the ground phorus supply on growth and phosphorus absorption. is moist and there is follow-up rain. We recommend the Aust. J. Agric. Res. 26: 423-36. best time for planting is in late autumn or early winter, Department of the Environment, Sport and Territories. though it can also be dry at this time of the year as experi- 1996. Refugia for biological diversity in arid and semi- enced in the 1999 trial. arid Australia. http://www.erin.gov.au/life/general_info/ biodivser_4/car.html. High on the priority list is a comprehensive seed collect- ing and storage programme of all species native to the is- Dixon, R 1999. Best management practices for the con- land, with key areas being the rich herbfields between the trol of perennial grass Ehrharta calycina. Manag- Acacias and other perennial plants when decommissioning ing our bushland. Proceedings of a conference about takes place. Some stabilisation trials also need to be un- the protection and management of urban bushland, pp. dertaken as cyclonic winds are a regular occurrence dur- 147-149. Urban Bushland Council W A Inc. ing the summer season. A recent cyclone altered the shape Griffin, G. F. 1993. The spread of buffel grass in inland of the island and caused erosion near a flare installation Australia: land use conflicts. Proceedings of the 10th which is being re-stabilised and revegetated by consult- Australian and 14th Asian-Pacific Weed Conference. ants based on the mainland. Brisbane, Australia, Sept. 1993; 10th Council of Aus- tralian Weed Science Societies. The ongoing success of the control of buffel grass and revegetation of Airlie Island with indigenous species de- Hacker, J. B. and Ratcliff, D. 1989. Seed dormancy and pends upon the good aegis of the funding sponsors who factors controlling dormancy breakdown in buffel grass manage the island. The programme does show that with accessions from contrasting provinces. Journal of Ap- careful, focussed research, it is possible to achieve effec- plied Ecology, 26: 201-212. tive and timely weed control in the arid zone. Hodgkinson, K. C.; Ludlow, M. M.; Mott, J. J. and Baruch, Z. 1989. Comparative responses of the savanna grasses ACKNOWLEDGMENTS Cenchrus ciliaris and Themeda triandra to defoliation; Oecologia 79 (1): 45-52. Western Mining Corporation (Petroleum Division) for ini- Humphreys, L. R. 1974. A guide to better pastures for the tial funding of the research project, Phase one and Novus tropics and sub-tropics. Wright Stevenson & Co. (Aust.) Petroleum for continuing this support. Apache Energy for Pty. Ltd. continuing funding and support of the implementation phase and for Airlie Island staff on ground assistance. Mr Humphries, S. E.; Groves, R. H. and Mitchell, D. S. 1991. Tom Vigilante, Project Officer for his input into part of Kowari 2 plant invasions. ANPWS, Canberra. Phase one of the project. Mr Geoff Kruger and Mr War- Parsons, J. M. 1995. (ed.). Australian weed control hand- ren Boggs, Department of Conservation and Land Man- book. Inkata Press. agement, Karratha, for their assistance with the on-ground spraying operations. WMC. 1993. Western Mining Corporation Airlie Island report second triennial report, 1993; Unpublished Re- port.

101